I have not yet made a working radiometer, but I have revived the project, crafting vane assemblies and wanting to quantify them to compare their relative masses.  After weighing one of my spindles on a kitchen scale, and then a postal scale, and finding them both reporting zero grams, I realized I needed something better.  I went to my favorite source of cheap but valuable scientific equipment, the UMN Re-Use Center, a place where the various university departments deliver their old-generation equipment that has been replaced by the latest technology.  The old technology is still amazing, and I have benefited from it, previously acquiring a high-power Ortholux microscope and other items for a fraction of their market value.

On this visit to the Re-Use center, I was looking for a laboratory triple beam balance, something I recalled from my high school chemistry classes, so I could record my various radiometer experiments.  There was nothing featured in their records, but I was invited to peruse their lab equipment and electronics sections.  I walked the aisles, not finding anything promising, and was about to leave when I encountered an object about 1-foot cubed, with a broken glass case, covered in dust, and stowed on the bottom shelf with other mystery equipment that had no apparent modern purpose.  

I did not recognize it, but it was an analytical balance, Model B120S, made by a company named Sartorius, perhaps used for years in a research lab or for chemistry classes until finally replaced.  I wondered if it could be resuscitated.  It was missing its power supply, but for $25 I decided I would take the chance.

I brought it home and had mixed feelings.  It was in terrible shape, corroded and encrusted by chemical debris, its draft-protection glass panels cracked. But I learned that this was originally a remarkable instrument, capable of weighing samples to a precision of four decimal places, a tenth of a milligram!  I couldn’t power it up to see if it worked, but I could clean it up.  I spent a weekend wiping and polishing parts, gradually disassembling the instrument to remove and dissolve the chemical residues of decades.

I also researched the past history of this instrument.  When I figured out how to access its working core, I learned from the date codes on the semiconductor chips that it had been manufactured in 1989.  This was a technology I was familiar with!

I needed to figure out the power supply for it.  I leaned on Google Gemini AI to help, and I got a series of encouraging responses and recommendations.  It was tempting to try them; I’m really glad I didn’t.

I know a former coworker who had moved to a company that worked with weights and scales.  I thought it was industrial level-  trucks and factory stuff, but the company he worked for turned out to be Sartorius Biolabs, which had products that covered the full range, including laboratory balances.  And sure enough, he was able to identify the power supply for the instrument I was restoring!  It had long since been obsoleted and unavailable, but it was nothing like the one that AI Gemini had suggested.  Had I followed its advice, I’d have likely destroyed the electronics.  Always remember the disclaimer:  AI responses may include mistakes.

I set about making a power supply replacement.  It involved three different voltages and a connector that was once ubiquitous in the 1970s:  the microphone jack.  You plugged it into your amp and you were rockin’.  In the 1980s, musical instruments were being connected to computers, and they used the same round barrel connector—MIDI.  Unfortunately, this common connector is almost, but not quite, entirely incompatible.  My connector had five pins. It had the right dimensions, but the pins were in different positions.  It would not mate.

The connector was also used in aviation and instrumentation of the time.  I had to locate a source for it that didn’t require me to buy a minimum of 100 (Digi-Key!).  Although it was once a very popular connector, it was backordered.  And I don’t think it was because they can’t keep up with demand. 

While I wait for that part, I will kludge something else up.  I can 3D-print a part that can hold wires in position to connect to the pins on the instrument.  Now I just needed to know what voltage went to what pin.

This turned out to be harder than expected.  I thought it would be documented somewhere in the seemingly unlimited resources of the internet, but this is an old instrument.  By the time the internet started accumulating PDF operating and service manuals for everything, this instrument was obsolete.  I managed to find a military manual for servicing a similar Sartorius scale, the A200S.  It gave me lots of information, but it did not have the pinout for the power supply.  Another valuable source came from a YouTube video by Jonny Fix, repairing a variation of that model.  It looked extremely similar to my B120S and used the same power supply!  Still, I could not glean the pinout.

So I re-entered the bowels of the unit and traced the wires.  Using an ohmmeter, I confirmed the routing between the connector pins and the internal voltage regulators.  I made a faux connector with a 3D printed core and some modified banana sockets, and wired it to my bench power supplies, which include a dual unit inherited from my dad’s ham radio shack.  Who knew it would end up here, providing a proof of function for an instrument that was almost the same age?  I eagerly awaited the moment of powering it up.  Would it be dead, and all this cleanup and wire-routing a waste of time?  Or would it come to life and demonstrate the amazing performance of this highly regarded laboratory workhorse?  Not that I would consider it, but a modern equivalent to this balance costs over $2000!

When my power supply and makeshift connector was finally ready, I had my moment of truth.  I turned on the power.  I was thrilled to see the display light up with the message “Power Off”.  This was infinitely better than seeing smoke come out of a circuit board component.  

I took the next step, pressing the ON/OFF button on the front panel.  The LCD display, a beautiful but delicate glass assembly, briefly flashed every segment it had, and then declared “busy”.  And it remained so.  I eventually turned off the power, relieved that there was still no smoke, but wondering what it was so busy doing, and how I would ever figure out what was needed to finish doing it.

I then noticed that one of the supply voltages was not keeping up- it was supposed to be 23 volts, but was only delivering 5.  I eventually figured out this was a safety feature of the old bench supply, to prevent a power overload.  The threshold had been set too low.  When I set it to a more realistic value, the unit happily provided the full 23 volts.

And now the startup sequence got further.  It was “busy” for a few seconds, flashed a few messages too fast for me to read, and then displayed a single large “L”.  I did not know what it meant, but it was listed in the military equipment manual as one of the symptoms of a broken machine.

I looked further into the error codes used by Sartorius scales.  In a manual for a much later model, I learned that “L” and “H” are used to indicate that the weight being measured is too low, or too high for the range of the instrument.  I could understand what too high meant, but what is too low?  Something below zero grams?

Yes, it was expecting the sample platter, a round metal disk with a center pin that directs its weight, plus that of the sample, to the measurement load cell.  I had not placed it into position and so it was far below the expected weight.

I put the platter in place, turned the instrument on, and was elated to see the momentary busy signal, followed by “0.0000 g”.  And further, when I blew gently at the platter, the numbers jumped around, and settled back to zero!  It was working!

Inspired by the demonstration by Jonny Fix (at the end of the YouTube link), I repeated it and noted that the weight of an ounce of water, about 30 grams, was not stable.  Amidst a bouncing value, presumably from air currents across the unshielded scale, there was a slow downward trend, a tenth-milligram every few seconds.  I looked up the evaporation rate of water from the surface of a shot glass, and it matched!  I was watching the loss of water molecules by evaporation.  At this rate, it would be 600,000 seconds until the glass was dry, 10000 minutes, 167 hours, or 7 days.  It seems plausible, but I didn’t run the test. I was just impressed that this instrument could detect the equivalent of paint drying, and maybe even the outgassing of my radiometer samples!

---

Gallery of photos documenting this tangent project. Click to enlarge, then scroll through. The explanatory captions are found at the bottom of the images.

---

A 2-minute video demonstrating the power up and first use of the scale.

---